Although there was no direct measure of learning for Pavlovian conditioning, all rats displayed characteristic behavioral effects of cocaine administration (e.g., increased locomotor behavior, rearing, stereotypy) after each training session. They were then trained to self-administer cocaine by performing a seeking-taking chain of actions. Over the course of training, rats showed a significant increase in both seeking and taking lever presses, with significantly more lever pressing on the seeking lever (see ). A repeated-measures ANOVA detected a significant main effect of day (F (1,14) = 35.284, p < 0.001), action (F (1,14) = 31.438, p < 0.001) and a significant interaction of day and action (F (1,14) = 5.601, p < 0.05). The difference in rate across actions is likely due to persistent responding on the seeking lever, which remained in the chamber during periods when the second (taking) component of the chain was active.
Acquisition of the seeking-taking chain, shown as average lever presses over the session for the last five sessions of training. Means + SEM. *** p < 0.001.
After Pavlovian and instrumental training, we tested the tendency for cocaine-paired cues to motivate lever-pressing behavior in the absence or presence of response-contingent cocaine deliveries. Pre-CS baselines are reported in . Surprisingly, a mixed ANOVA conducted on the pre-CS baselines revealed a significant main effect of CS (F (1,13) = 6.19, p < 0.05), and a significant action by group interaction (F (1,13) = 5.89, p < 0.05), with no other significant main effects or interactions (largest F value: F (1,13) = 4.26, p > 0.05). These results indicate higher baselines for the CS− than the CS+, and more responding in the baseline period on the taking lever for the extinction group. Since these cues were delivered in strict alternation, the relatively elevated baseline response rate going into CS− trials may reflect carry over of the excitatory influence of recent CS+ deliveries.
Test 1. Pre-CS baselines and extinction responding
The results of the first PIT test are presented in . Although the findings from this initial round of testing were not particularly clear, some features of these data are worth noting. First, it appears that rats tested in extinction show a slight increase in lever pressing during cue presentations, an effect that was at least numerically greater during CS+ trials and was most prominent once the cues were terminated. Second, for rats that were rewarded with cocaine at test, there was little indication that the CS+ was capable of invigorating cocaine seeking or taking behavior. An ANOVA performed on these data found no effect of CS (F (1,13) = 1.958, p > 0.05) or action (F (1,13) = 1.396, p > 0.05), nor was there a significant interaction between these variables (F(1,13) = 0.177, p > 0.05), or between these variables and test group (F = 0.03, p > 0.05). However, though not specific to action or CS, there was a significant effect of period (F (2,26) = 4.907, p < 0.05) and a period by group interaction (F (2,26) = 4.365, p < 0.05), indicating that the groups differed in the rate at which they lever pressed during and immediately after the cue presentations. To explore this effect further we conducted separate CS x Action x Period ANOVAs for each test group. Both groups demonstrated a significant main effect of period (F (2,12) = 4.23, p < 0.05 – extinction group, F (2,14) = 3.90, p < 0.05 – rewarded group), with no other significant main effect or interactions (largest F value: F (2,14)=2.741, p > 0.05).
Figure 2 Results of the first Pavlovian-to-instrumental transfer test. Difference scores for each minute of the CS and the first minute of the post-CS period, displayed separately for action (seeking or taking) and group (extinction or rewarded). A–B, (more ...)
In light of these results, we conducted a second test using procedures likely to facilitate the expression of PIT. First, given the clear lack of effect in the rewarded group, all rats were tested under extinction conditions in Test 2. Second, we increased the initial extinction period from 5 to 10 minutes to further suppress pre-CS response rates and avoid a potential behavioral “ceiling” or upper limit on responding. Third, we increased the interval between trials to 6 minutes to minimize carryover effects. Fourth, we gave rats 3 sessions of extinction of the CS+ before Test 2. Though counterintuitive, recent evidence suggests that such extinction of the CS+ can enhance the PIT effect, presumably by weakening competing conditioned responses (Holmes et al., 2010
Pre-CS baselines for Test 2 are reported in . There was a significant main effect of action (F
(1,13) = 7.05, p
< 0.05), but no other significant effects (largest F value: F
(1,13) = 2.17, p
> 0.05), reflecting more pre-CS responding on the taking lever than the seeking lever. presents the results of the Test 2. As is clear from these data, rats displayed a stimulus-specific increase in lever pressing to the CS+, a pattern indicative of PIT. A repeated-measures ANOVA showed a significant main effect of CS (F
(1,13) = 6.752, p
< 0.05), indicating that the enhancement in lever pressing was greater for the CS+ than for the CS−. There was no effect of action (F
(1,13) = 0.324, p
> 0.05) or period (F
(1,13) = 0.208, p
> 0.05), nor were there any significant interactions between any of these factors (largest F value: F
(1,13) = 2.25, p
> 0.05). Although these analyses failed to identify action-specificity in the PIT effect, inspection of the data in suggests that the CS+ produced a more pronounced and longer-lasting enhancement in performance of the taking action than of the seeking action, consistent with previous food PIT studies (Corbit & Balleine, 2003
). Indeed, when we confined our analysis to one or the other action, we found a significant effect of CS for the taking lever (F
(1,13) = 5.795, p
<0.05), but found no such effect for the seeking lever (F
(1,13) = 1.68, p
Results of the second Pavlovian-to-instrumental transfer test. Difference scores for each minute of the CS and the first minute of the post-CS period, separately plotted for seeking (A) and taking (B) levers. Means + SEM.
In view of the trend towards a difference in the degree of transfer on the seeking and taking levers, an additional analysis of transfer test performance was conducted to determine if the rats performed the chain task during the test as trained, i.e. shifting from the seeking lever to the taking lever rather than moving in the opposite direction. To quantify these shifts in performance we followed Balleine and colleagues’ example (Corbit & Balleine, 2003
) in assessing the probability that the taking lever action was performed in each of the 10 seconds that followed performance of the seeking action, and generated a similar probability distribution for seeking presses that occurred in the 10 seconds that followed each taking action. These data, which are shown in , clearly demonstrate that the rats were more likely to shift from seeking-to-taking, which was the order of actions reinforced by cocaine delivery, than from taking-to-seeking. This shift in performance appeared to be most frequent during approximately the first five seconds in the post-seeking period. A repeated measures ANOVA using action order (Seek-Take vs. Take-Seek) and time bin (1–10) as factors confirmed this analysis, revealing a significant main effect of action order (F
(1,14) = 22.121, p<0.001), a significant main effect of bin (F
(9,126) = 8.176, p<0.001), and a significant interaction between action order and bin (F
(9,126) = 8.029, p<0.001).
Figure 4 Probability of transitioning from the seeking to the taking lever vs. transitioning from the taking to the seeking lever. Probabilities are calculated by dividing the total number of transitions in each 1-second bin by the total number of 1st action lever (more ...)